OMP: One-step Meanflow Policy with Directional Alignment

TL;DR

The paper introduces OMP, a novel one-step meanflow policy for robot manipulation that achieves high-fidelity, real-time control by addressing spectral bias and gradient issues, outperforming existing methods on benchmarks.

Contribution

The paper proposes OMP with a directional alignment mechanism and a DDE for efficient, high-precision robotic manipulation, overcoming theoretical limitations of existing meanflow approaches.

Findings

OMP outperforms state-of-the-art methods in success rate.

OMP achieves higher trajectory accuracy in high-precision tasks.

OMP maintains efficiency with single-step inference.

Abstract

Robot manipulation has increasingly adopted data-driven generative policy frameworks, yet the field faces a persistent trade-off: diffusion models suffer from high inference latency, while flow-based methods often require complex architectural constraints. Although in image generation domain, the MeanFlow paradigm offers a path to single-step inference, its direct application to robotics is impeded by critical theoretical pathologies, specifically spectral bias and gradient starvation in low-velocity regimes. To overcome these limitations, we propose the One-step MeanFlow Policy (OMP), a novel framework designed for high-fidelity, real-time manipulation. We introduce a lightweight directional alignment mechanism to explicitly synchronize predicted velocities with true mean velocities. Furthermore, we implement a Differential Derivation Equation (DDE) to approximate the Jacobian-Vector Product (JVP) operator, which decouples forward and backward passes to significantly reduce memory complexity. Extensive experiments on the Adroit and Meta-World benchmarks demonstrate that OMP outperforms state-of-the-art methods in success rate and trajectory accuracy, particularly in high-precision tasks, while retaining the efficiency of single-step generation.